Football Outsiders content published by ESPN
PDF VERSION NOW DISCOUNTED OVER 30%
Click here to buy PDF version.
Click here to buy PDF version
Official Account: @fboutsiders
Scott Kacsmar: @FO_ScottKacsmar
Ben Muth: @FO_WordofMuth
Aaron Schatz: @FO_ASchatz
Vince Verhei: @FO_VVerhei
-- plus --
Bill Connelly: @SBN_BillC
J.J. Cooper: @jjcoop36
Cian Fahey: @Cianaf
Brian Fremeau: @bcfremeau
Tom Gower: @ThomasGower
Andrew Healy: @AndHealy
Rivers McCown: @RiversMcCown
Chad Peltier: @CGPeltier
Matt Waldman: @MattWaldman
Rob Weintraub: @robwein
10 Nov 2008
Where does Jake Delhomme rank among the worst quarterbacks in the DVOA era to actually win a game?
Posted by: Bill Barnwell on 10 Nov 2008
48 comments, Last at
12 Nov 2008, 1:24pm by
White Rose Duelist
Through 9 games, Peyton Manning has faced 5 of the top 11 defenses in terms of defensive passer rating. In his last 7 games, he'll face 5 of the 10 worst defenses in terms of passer rating.
Don't be surprised if he winds up with 30-35 TDs and a strong MVP bid. People will come up with all kinds of reasons for the turn around, but most of it will just be facing an easier set of defenses.
Yes, pundits have been pulling the "offense is just out of sync" card for a while. And of course the Colts will pull it together in the second half of the season when supposedly Manning's knee finally heals and he has had 10 weeks to practice with his receivers (yeah, it only takes 10 weeks for a QB of Manning's caliber). I still contend their largest liability all year has been the defense (big surprise!). They cannot get off the field. There are times I yell, "If your gonna let them score, let them score fast! Just stand there, let them score the touchdown and get the offense back on the field!"
On a side note, these CAPTCHAs are making me feel pretty stupid. And it is an acronym - Completely Automated Public Turing test to tell Computers and Humans Apart. Scary times we live in, scary times.
Or put in terms of DVOA...Manning has faced 6 top 10 passing defenses in 9 games. Now he'll face 6 of the bottom 10 in 7 games. And his other game...a possibly meaningless matchup for Tennessee.
"Chicago runs less Cover 2 now than it ever has under Lovie Smith, however, and Tennessee adapted by challenging Chicago's corners in man coverage with lots of slants. The corners couldn't handle it."
Really? At the game, it seemed like the Bears were playing the same old Cover 2, with soft zones by the corners. Maybe it was just really soft man coverage, but does that really count as man coverage? Isn't that just zone coverage?
It almost all cover 3 and cover 1. The point of cover 2 is that the corners can play short routes aggressively.
Bears corners don't play short routes aggressively. They play outside leverage, giving the other team as many slants as they want.
What would Cassel's numbers have been if you took away those pathetic sneak "attempts" at the end of NE's final drive (despite what Belichick says, I have my doubts they were trying very hard on those sneaks -- I think they were trying to run the clock down before either scoring or pinning Buffalo on its own 1)?
Also, where did Welker finish up on the day? He caught 10 balls and to my memory he got a bunch of first downs and snagged most of what was thrown to him.
How the hell did asomugha get drafted 27th overall that year?
He was actually 31st overall that year, and while you could make arguments for a couple handfulls of the guys in front of him (Carson Palmer, Troy Polamalu, etc), he might be the steal of the draft--definitely the steal of the round, and possibly, the draft (although Anquan Boldin, Osi Umenyiora, Lance Briggs, and Jason Witten may have something to say about that)
Are you really complaining that a player went in the first round? Plus, he was only the 5th corner taken, and two of the corners taken before him, Terrence Newman and Marcus Trufant, aren't exactly scrubs.
Quite a few good players were drafted before him, and I think several of them would still go before him if they redid the draft. Not every team wanted a corner.
He was a beast of a safety, but he was still a safety. It wasn't clear by any means that he'd be able to develop the coverage skills of an elite corner.
And, honestly, he just wasn't that brilliant when he was drafted. Some people get better as their careers go on. They aren't total knockouts in college, but years of hard work bring them to the top. Think Donald Driver, Bob Sanders, Aaron Kampman, or Tom Brady.
Bob Sanders was phenomenal in college. He wasn't a higher pick because of his size and injury history (only one of those turned out to be an issue, (or they are actually the same issue)).
Does anyone really believe that Ricky Williams was more valuable than Adrian Peterson this week?
I don't want to debate the merits of DYAR or effective yards I'm just asking does anyone really believe R. Williams was more valuable.
No one does - just like no one thinks Peterson is the 17th best starting back in the NFL as rated here on this site.
I think this site does an excellent job on rating teams and various aspects of the game (rushing, passing, special teams) but I find the individual ratings almost pointless and the example above is about as pointless as it gets.
Imagine that Green Bay had a competent offense and drove for a TD after Peterson's fumble. The game is 31-21 with four minutes (max) to go. Minnesota needs to score twice, so Frerotte has to throw -- probably no late, great run from AP. The Packers probably win.
Sure, Peterson is more valuable to his team than Williams, because the Vikes have no pass offense. But without that defensive stand (and the subsequent one, which forced the missed FG) the postgame talk would be about Peterson's fumble, Frerotte's uselessness, and Childress's inability to call his own plays on fourth down.
Well he is Boom/Bust-y - and FO prefers the more consistent backs.
That said, I agree on the case in hand. I think DVOA, somehow, doesn't take into account, that AP carries that ofense on his back - ALONE! And that his main assignment is to keep the clock moving and the ball out of Frerottes hands.
You guys realize as is said AGAIN and AGAIN, that the individual rankings are not Williams vs Peterson.
They are Williams running the ball in the Miami Offense (an offense which has a solid passing game) vs Peterson running the ball in the Minnesota Offense (an offense which has a horrible passing game).
No one at FO is claiming the stats say Williams is better than Peterson (sometimes they do slip and make this mistake in their articles, possibly because it makes them sound more racy).
They are claiming that Williams is more valuable than Peterson when combined with his overall offensive environment. I am guessing Peterson faces a lot more 8 men in the box or 7 men in the box and another two staring into the backfield each snap than Williams does.
As of now DVOA doesn't take into account how defenses adjust to different offenses. It is a limitation you need to recognize and keep in mind.
"Love is like a bottle of gin (not verified)"
Actually I can verify that.
"I think this site does an excellent job on rating teams and various aspects of the game (rushing, passing, special teams) but I find the individual ratings almost pointless and the example above is about as pointless as it ge"
Completely, 100% agree. The individual rankings are misleading.
It's mainly because the "individual rankings" aren't really individual rankings...they're rankings of specific aspects of a team (something that most FO readers seem to forget).
The ratings this week don't say that Peyton Manning was the most valuable QB...they say, that the Colts passing attack (led by Peyton Manning) was the most valuable.
Steve Smith wasn't the least valuable wide reciever...the portion of the Panther's offense they tried to direct towards Steve Smith was the least effective.
"Keller is a very good all-around tight end, and he will be starting for the team on a permanent basis any day now."
Keller is nowhere near the blocker that Baker is, which is why Baker is still starting. Baker also has better hands, but Keller is much faster.
3 Bear's games on the worst winning QB ever list. I don't know whether to be embarrassed or proud.
You took the words right out of my mouth.
I'm almost surprised all 5 of those games weren't Bears games.
I've completely blocked it out of my mind, now that we have some semblence of a franchise quarterback. But yes, that makes the "There is only 'Chicago QB' " jokes of old even more funny.
Hate to break it to you man, but Hester's not a franchaise QB.
Shockingly, none of them were by Orton in his rookie year. I thought for sure he would make that list, as the "game manager" who just barely managed to do little enough damage that his team still won on quite a few occasions.
(Formerly "The McNabb Bowl Game Anomaly")
The 2005 Bears didn't have Hester, so if Orton blew up, they could only score on defense. At least partially as a result of that, the Bears actually lost the games in which Orton was truly bad (such as the Cincinnati game when he threw at least 5 picks). To his credit, Orton was (for the most part) consistently mediocre.
Quick correction: Garrard and the Jags played the Lions, not the Rams.
Ammek - Peterson was hit short of the first down and it seemed very unlikely he was going to make it. The fumble was irrelevant if he didn't get the first down. And the notion that a back successfully making a 3rd or 4th and short is mostly about his ability is beyond ridiculous.
I agree. But since most of the post-game AP-love has centered on that final drive (although he played well pretty much all game -- without much help), I wanted to emphasize how it nearly did not happen at all; unfair as it may be, I'm sure you'd agree that, had the Packers won 31-28 as in my example, the post-game angle would have been: "Minnesota finds yet another way to lose a game it dominated", in which the 4th-quarter challenge/4th down call/failure/fumble would feature prominently.
Well, that Eli Manning bias once again comes through.
Okay, I gotta ask: How can numbers be biased?!
I was referring to the text, not the numbers. The Giants were down in the second half, you expect them to run constantly? And you expect Eli to go against Couglin and Gilbride's playcalling for passing? Blaming Eli for passing instead of running the balling despite completing 12 passes for 120 yds with 2 tds and 1 int before McNabb completed a pass? The game was close because of stupid turnovers put into tds by the Eagles, but I really find it funny how much this site hates Eli Manning.
Manning was good enough, but every time he dropped back, all we could think was "Your running backs are getting 7 yards a pop. Just hand them the ball, Eli."
I don't think this comment was suggesting that Eli audible away from Gilbride's calls. I think it was really aimed at Gilbride himself. But what do I know.
I find it really funny how fans of certain teams inevitably delude themselves into thinking that the writers of this site are biased, when they have noted on numerous occasions that Eli is having the best year of his career.
On worst winning QBs, where does David Carr rank from Houston vs. Steelers in 2002? 3-10, 33 yards (all to Billy Miller), four sacks, two fumbles, (Texans won 24-6 off two Tommy Maddox pick-sixes, a fumble returned for a touchdown on a play Maddox wasn't hit, and a field goal off a fumbled punt.)
Does he escape by opponent adjustment, or because DYAR is cumulative and he didn't pass that much? Please God, is that the worst game by a QB who won by eighteen points?
That game was freaking hilarious.
Brown 10 carries for 39 yards 1 TD throw out HIS long TD run and you get 9 carries 23 yrds. Williams 11 carries 54 yards. So how exactly are they pretty much the same? Peterson had a fumble (they recovered it), but football outsiders don't care in their system since fumble recovery is random.
I think (and I'm speculating here) the point is that Williams' long run could have easily been done by Brown instead, being that he was untouched and just had to run fast in a straight line. So apart from that, they were roughly equal.
Also, Football Outsiders doesn't care who recovers the fumble. They definitely care who had the fumble. All fumbles are equally bad from a process standpoint, even though only lost fumbles hurt the team's outcome. (FO measures process, not outcome.)
How does Grossman nearly crack the top 10 this week? It must be opponent adjustments. He was awful, missing wide open receivers, throwing into coverage, completed less than 55% of his passes, not to mention third down conversion (3 of 14 - both running and passing, but still).
As great as his early game scrambling went, he ended up as just-below average on the day because of the two failed sneak attempts he went on in the fourth quarter.
This highlights what I see as the biggest remaining deficiency in FO's stats--aside from Aaron's manual efforts to throw out "garbage time" stats, DVOA and DYAR rate players assuming a game is infinitely long.
There are times when a zero-yard run is actually more useful than a 3 yard pass that stops the clock, or even than a TD, but DVOA and DYAR miss that. I think having 1st and goal from the 1, when the other team is out of timeouts and you're close to the two-minute warning, is exactly one of those times. It's the same reason Mike McCarthy's challenge in the Packers-Vikings game this week was so stupid...if you're the defense, you're pretty sure the other team is going to score (unless, obviously, you're facing the 49ers), so, if you're already trailing, you want them to score as quickly as possible so you can get the ball back.
It was pretty clear from the half-hearted blocking and the lack of effort from Cassel that those two sneaks were intended to end the play with as little loss as possible and run as much time off the clock as possible, without getting into the endzone. So by the time the Pats did score, the Bills had way less time than they would have if the Pats had scored right away (they would have had 4:00+, if I remember correctly).
People always complain when folks bring up problems with FO's stats without making suggestions on how to fix them, so here's my suggestion. Change the way "success points" are calculated to include a dependence on time consumed by the play. Count making the other team burn a timeout as as, say, a 35 second loss (since a team with a timeout can save themselves about 30 seconds if they have the ball and 40 seconds if the other team does). Make the "success" value of time depend on whether you're leading or trailing, and how close it is to the end of the half/game. So for most of each half, time consumed by a play would not affect DVOA/DYAR ratings. But near the end of each half, you penalize teams that are trailing and take up lots of time, and benefit teams that are leading and eat clock. To a trailing team under 2:00, a 20 yard sideline pass that gets out of bounds should be worth more than a 35 yard seam pass to the middle of the field.
You'd have to play with the function that weights the time value, in order to optimally correlate (the same way everything else on FO has been tuned), but it could be very valuable. And, it would allow you to start using FO stats to evaluate the value of challenges, timeouts, personnell subs that force a timeout, etc., and compare them to play outcomes in the same terms.
You make good theoretical points that merit some investigation. However, using a system that DOES account for time left (and is very exhaustive, I might add), up 13-3, the Pats had already essentially sealed up the game after this play:
(3:26) 42-B.Green-Ellis left tackle to BUF 1 for 2 yards (91-Sp.Johnson, 92-R.Denney). PENALTY on BUF-99-M.Stroud, Defensive Holding, 2 yards, enforced at BUF 3 - No Play.
At this point, the Pats had nearly a 100% chance to win the game (the two Cassel plays came after). So while getting no gain is probably a higher-value play, there was no MARGINAL benefit to them. So really, while they shouldn't be marked as "failed" plays, they shouldn't be marked "successful" either. Running off time at that point does not improve your chance of winning very much... even if it is the correct strategy.
EDIT: Sorry, forgot to link my source for the marginal benefit: http://18.104.22.168/finalgames.php
As you can see, after they get down to about the 20, the probability of winning the game remains about the same until the holding penalty, after which they have just about 100% chance of winning.
EDIT2: Actually, now that I think about it, this only measures the probability of winning given the actions on the field. It does not show how the probability of scoring would be (would we see a small swing in Buffalo's favor if the Pats score?). Eh, whatever, I'll think about this later.
I think you have some really good ideas, but I'm not sure if they can be applied in a useful way. Consuming or saving time at the end of a close game is critical (and easily measured by play-by-play), depending on if you're leading or losing. At the end of a half, I think it's trickier, because even a trailing team will prefer to use up time so the other team doesn't get another chance to score. Putting these measures in at the end of the half might be impossible. Even at the end of the game, a team trailing by 3 or less might try to run out most of the clock so they can end the game on the FG attempt. Similarly, teams which are tied will try to save just enough time to attempt to win but use up the rest.
I'd love to see Aaron try something, but I suspect it won't be workable in the end.
Agreed--I don't think it's workable for the computers, either. A good example would be the end of the Penn St./Iowa game this Sat. Iowa is trailing, but let the clock go down to six seconds so that they can (presumably) kick the game-winning FG without leaving Penn St. the opportunity to drive back down the field for their own GW FG. NFL teams have done similar things also. A computer might penalized the trailing team for taking extra time--when it is the correct thing to do.
As for the Cassel "failed" sneaks, I think the thing would be for FO to remove them in the same way they remove end of half/game kneel-downs. I didn't see them, but it seems like from the comments that they were basically kneel-downs, although not actually done that way.
DVOA does use the time remaining in the game when determining the value of plays. A 1-yard run on 1st-and-10 is worth more when the team with the ball is up by inside the two-minute warning than during the first quarter. I don't know how much difference it makes in this case; perhaps it would be treated differently if Cassel just knelt ont he ball twice rather than "attempting" to score. However, two failed sneaks would be penalized more earlier in the game.
If you remove those two "failed sneaks"--which weren't so much failed sneaks as "let's burn another minute or so off the clock" plays--what happens to his rushing DYAR?
I just look at Matt Ryan's numbers after every week, and marvel that this is a rookie quarterback. I know he's faced some really bad defenses, but still. What will this guy do when he's been in the league for awhile and has had a few years to work with Roddy White?
The Vikings need offensive line help, while the Bears, Lions, and Packers have significant defensive concerns.
See All XP | NFL XP | College XP
© Football Outsiders, Inc. // Site powered by Stein-Wein // Partner of USA TODAY Sports Digital Properties